Yuma has accelerated its own subnet: Loosh (SN78)
The launch of Loosh (SN78) marks the introduction of a Yuma-accelerated subnet designed to push machine intelligence into more human-aligned territory. The project focuses on developing cognitive frameworks, ethical reasoning, and emotional intelligence for autonomous agents and robots – capabilities considered essential for building systems that act consistently and can earn user trust.
What is Yuma?
Yuma is a decentralized AI accelerator and infrastructure builder operating within the Bittensor ecosystem. It supports and scales early-stage research teams and entrepreneurs by providing access to capital, technical resources, and community engagement to help launch and grow subnets on Bittensor. Yuma’s mission is to expand access to open, decentralized intelligence as an alternative to closed AI systems and to foster innovation across the network.
What is Loosh (SN78)?
Loosh is an AI subnet and consciousness framework. It aims to map and simulate human awareness by turning cognitive and nonverbal human states into structured, machine-readable data. Through purpose-built subnets, Loosh seeks to give robots and agents persistent memory, long-term reasoning, emotional understanding, and context awareness – effectively engineering a scalable infrastructure for machine consciousness that blends neuroscience, data science, and decentralized design.
Our goal is to deliver Cognition as a Service through scalable, API-accessible infrastructure built on Bittensor. The platform will support applications across telehealth, neurodivergent communication, ethical robotics, and adaptive digital companions. Loosh agents will be capable of emotionally-aware interaction, continuous self-refinement, and ethical decision-making aligned with human oversight and societal norms.
How does Loosh work?
Loosh is building a system that allows AI agents to develop more human-like understanding, context awareness, and emotional grounding. It does this by combining three main components:
Loosh defines a structured model of machine consciousness built from multiple interconnected functions: perception, memory, emotional interpretation, ethical reasoning, and long-term planning.
Instead of relying on one large model, Loosh breaks intelligence into smaller parts that work together to interpret situations, understand intent, and make more trustworthy decisions.
The project uses tools that help researchers study how human experiences and mental states relate to measurable data. This includes collecting and visualizing information about emotions, attention, and subjective experiences, then using those insights to refine how Loosh models cognition.
In practice, this research guides Loosh in shaping AI behaviors that better reflect human values and patterns.
Loosh uses a vector-based memory system that lets agents store and retrieve past experiences, contextual cues, and emotional signals. This enables continuity: agents can remember relevant information over time instead of treating every interaction as isolated.
